BACKGROUND:

The diagnosis of acute pancreatitis relies heavily on a raised amylase.

METHODS:

In the present study patients were prospectively categorized, without knowledge of pancreatic enzyme levels, into acute pancreatitis (AP; n = 51), disease controls (n = 35), indeterminate as to pancreatitis (n = 189) or exclusions (non-pancreatitis diseases where amylase may be elevated; n = 53).

RESULTS:

Enzyme levels were analysed by receiver operator characteristics (ROC) curves, with specificity > 80%. Day 1 serum lipase gave the greatest diagnostic accuracy (area under ROC curve = 0.128; P = 0.041 vs serum amylase). At the calculated diagnostic threshold of 208 U/L, lipase gave a sensitivity of 67% and a specificity of 97%. Other diagnostic thresholds (day 1) were: serum total amylase, 176 U/L (ROC 0.104, sensitivity 45%, specificity 97%), urinary total amylase, 550 U/L (ROC 0.108, sensitivity 62%, specificity 97%) and serum pancreatic isoamylase, 41 U/L (ROC 0.107, sensitivity 63%, specificity 85%). At delayed diagnosis (3 days) no enzyme was superior to lipase. The combination of lipase and amylase did not increase diagnostic accuracy. Conclusion: Serum lipase is recommended for diagnosis of AP, both early and late in the disease. Although highly specific when elevated, all pancreatic enzymes have low sensitivity for diagnosis.